Study On Vegetation And Building Height Inversion With PolINSAR And TomoSAR

Polarimetric Interferometry Synthetic Aperture Radar(PolInSAR)technology includes not only interferometric phase information being sensitive to elevation information of ground objects,but also polarization scattering information being sensitive to the shape,material,structure,direction and spatial distribution of natural objects.It has great potential in the fields of vegetation height inversion,high precision topographic mapping,efficient forestry resource management,agricultural production,meteorological mapping,disaster prevention and mitigation,accident search and rescue,anti-terrorism reconnaissance and so on.Tomographic Synthetic Aperture Radar(TomoSAR)is a technique in which InSAR system gradually increases the number of baselines along the elevation to form synthetic aperture,and then obtains the high resolution of ground objects through high-dimensional imaging processing.This technology has been widely used in building height inversion,building three-dimensional structure extraction,building micro-deformation monitoring and vegetation vertical structure inversion.In this dissertation,the key problems of vegetation height inversion based on PolInSAR and building height inversion based on TomoSAR are studied.The vegetation height in this dissertation refers to the general vegetation height,including the digital surface model(DSM)of vegetation area and the vegetation canopy height from the ground surface to the top of vegetation.This dissertation mainly studies the theoretical basis and data preprocessing of PolInSAR,DSM estimation of vegetation area by PolInSAR,inversion of vegetation canopy height under X-band PolInSAR and inversion of building height by TomoSAR.The main contents and innovations are summarized as follows:(1)PolInSAR Theory and Data PreprocessingIn the second chapter,the basic theory of PolInSAR is briefly introduced.The speckle noise filtering methods and the coherent coefficient estimation method of PolInSAR in sparse vegetation area are studied.The main work is as follows:A joint restrict principle for speckle noise filtering of polarimetric SAR is proposed.In order to overcome the shortcomings of traditional polarimetric SAR speckle filters in maintaining three kinds of performance,including the edge texture,the intensity statistics and the polarimetric scattering characteristics,a morphological window which can automatically match the arc or linear edges of the ground objects and a polarization Scattering Similarity Factor(SSF)which can maintain the polarization scattering characteristics of the ground objects are proposed,and an extended Lee-sigma filter which preserves the statistical characteristics of image intensity is introduced.A polarimetric SAR speckle noise filtering method combined with the above three constraints pricinples is proposed.Firstly,the sample pixels which preserve the texture characteristics of the image are selected by the morphological consistency principle.Then,the selected sample pixels are filtered again by the other two principles and the qualified sample pixels are retained.Finally,the noise filtering is performed.After filtering,the above three performances of the image can be maintained at the same time.The effectiveness of the proposed method is verified by the real data of spaceborne polarimetric SAR.A method for estimating PolInSAR coherence coefficients in sparse vegetation areas is presented.Without considering the effect of interstice in the traditional estimation of sparse vegetation regional coherence coefficient causes an inaccurate estimating results of coherence coefficient.The joint constraint criterion is applied to the estimation of polarimetric interferometric coherence coefficients,and a method which estimates both the self-correlation matrix and the cross correlation matrix of the master and slave tracks based on the joint constraint principle,so as to obtain the accurate estimation result of the coherence coefficient is presented.The validity of the method is verified by the airborne real data.(2)DSM Estimation of PolInSAR Vegetation AreaChapter 3 focuses on polarimetric interferometry coherence optimization technology and high-precision DSM estimation method for PolInSAR vegetation area based on this technology.The main contents include:A method of maximum phase difference optimization for DSM estimation in PolInSAR vegetation area is presented,and a method of DSM estimation in PolInSAR vegetation area based on combined interferometric coherence coefficient amplitude difference and interferometric coherence coefficient phase difference optimization is introduced.The former maximizes the height of scattering phase center of electromagnetic wave in vegetation by solving tangent value of coherence coefficient.The latter considers the optimal correlation between the amplitude difference of the interferometric coherence coefficient and the phase difference of the interferometric coherence coefficient,and takes into account the improvement of both the quality of the polarimetric interferometry phase and the height of the scattering phase center of the electromagnetic wave in vegetation.Considering that DSM is sensitive to the phase of interferometric coherence coefficient,when the amplitude of interferometric coherence coefficients is high after the coherence optimization,the maximum phase difference optimization method is the priority to estimate the final DSM.This dissertation presents a DSM estimation method for PolInSAR vegetation area based on the estimation of the least penetration depth of electromagnetic wave into vegetation.In view of the penetration depth of electromagnetic wave in vegetation,there is still a certain distance between the results of DSM estimated by maximum phase difference optimization method and the real vegetation top elevation,that is,DSM is still underestimated.In this dissertation,an exponential model for estimating the least penetration depth of electromagnetic wave in vegetation is proposed.For DSM estimation,the interferometric phase of high scattering phase center obtained by maximizing phase difference optimization method is used to obtain the rough surface elevation of vegetation area,and then the least penetration depth of electromagnetic wave is compensated to the rough surface elevation,so as to improve the accuracy of DSM estimation in vegetation area.The validity of the proposed method is verified by simulation data and airborne real data.(3)X-band PolInSAR Vegetation Canopy Height InversionChapter 4 studies the inversion technology of vegetation canopy height of PolInSAR in X-band.The main contents are as follows:This dissertation proposes an X-band PolInSAR vegetation canopy height inversion strategy based on Frequency Segmentation(FS)of SAR images with different polarization channels.In order to solve the problem of inaccurate inversion of vegetation canopy height caused by large errors of terrain phase estimation and effective volume scattering coherence coefficient estimation in traditional three-stage inversion algorithm due to the too concentrated distribution of coherence coefficient in X-band,FS is applied to extend the range of the interferometric coherence coefficients of the original SAR images in different polarization states to obtain the precise terrain phase.For the selection of effective volume coherence coefficient,part of the straight line which is used to fit the extended range of coherence coefficient distribution is intercepted by the fixed range extinction curve to get the accurate effective volume coherence coefficient.Finally,the high-precision vegetation canopy height in X-band is estimated by the effective volume coherence coefficient which removes the terrain phase.The validity of the proposed method is verified by the airborne real data.(4)TomoSAR Building Height InversionIn the fifth chapter,the baseline optimization design of TomoSAR imaging and building height estimation using TomoSAR technology are studied.The main work is as follows:A baseline optimization method for Iterative Space-perturbation on Sequential Quadratic Programming(ISP-SQP)is proposed.Aiming at solving the problem that the traditional baseline optimization method is easy to converge to local optimum when building height inversion is carried out by tomography method,the spatial disturbance is added to the baseline distribution iteratively based on the traditional baseline optimization results,and the global optimum baseline design results are obtained finally.The feasibility of the proposed method is verified by simulation data.A new inconsistency criterion based on Joint Phase and Amplitude(JPA)is proposed.Aiming at solving the diffusive problem of homogeneous points in high resolution SAR images with different tracks,which affects building height inversion,a criterion of inconsistency of pixel amplitude and phase is constructed.The criterion uses the inconsistency of neighborhood pixels amplitude and phase in the reference pixel window to judge homogeneous pixels in different SAR images.By using this criterion,the homogeneous points diffusing under different track conditions can be found effectively,thus improving the accuracy of TomoSAR building height inversion.The effectiveness of this method is verified by using simulation data and airborne real data.A method of building height inversion based on Singular Value Decomposition(SVD)is presented.The traditional SVD method only deals with the measurement matrix.The large phase error of the observation vector in low signal-to-noise ratio(SNR)leads to the large tomography error.A new method of SVD tomography combined with signal subspace is proposed.This method first introduces an M estimator without considering the noise distribution form to estimate the covariance matrix of the observed signal,then decomposes the covariance matrix by eigenvalue decomposition.The eigenvectors under the order of the model are selected to form the signal subspace,and the denoising observation vectors are accumulated by the signal subspace.Finally,the observation vector is applied to the SVD method to improve the accuracy of building height inversion.The validity of the algorithm is verified by simulation data and airborne real data.